The radiative transfer code is adapted from the one
developed in the context of terrestrial modeling
by Fouquart and Bonnel (1980) for
solar radiation and Morcrette et al. (1986)
for thermal radiation. This code, originally developed for the climate
GCM of Laboratoire de Météorologie Dynamique, has since been
included by Morcrette in the operational model of the European Center for
Medium-Range Weather Forecasts (ECMWF).
Recently, at LMD, in the frame of extension to ESA Contract
11369/95/NL/JG, the infrared part of the radiative transfer
code has been rewritten with a net exchange formulation.
effects of both atmospheric carbon dioxide and
dust are included.
Particular attention was given
to the parameterization of absorption by the
m band with inclusion of Doppler effect [3].
Cooling rates computations were carefully validated by
comparison to line-by-line integrations. The model, in which Doppler
effects are introduced, is very accurate up to 70 km.
The thermal spectrum is divided into three parts, one for the core of
the
m band, one for the wings and the third one for
the remaining part of the spectrum.
For the three parts, the
transmission by dust is computed using grey absorption approximation.
For the two intervals of the
m band,
the total transmissivity is evaluated as the product between
transmissivity of dust and that of carbon dioxide.
Strictly speaking, it can be shown that
this evaluation of the combined transmissivity is valid
when there is no correlation between the spectral variations of the
two absorber. This is generally assumed for dust and
carbon dioxide in the
m band.
Scattering is not taken into account because of the strong isotropy of
the thermal radiation.
This new code is based on a net exchange formulation.
The original Morcrette radiative code computes directly fluxes at each
level.
In the new approach, we first compute individual exchanges for each pair
of layers,
and then the budget for each layer.
The net exchange between two layers is the product of two terms:
1) the difference between Planck functions, which depends on temperature
profile,
2) the exchange coefficient, which only depends on optical properties.
The main interest of this new method is
first to allow analysis of the relative importance of
the various terms included in the budget (in the flux formulation, we
just have a global value for the exchange of the considered layer with all
the rest of the atmosphere), and second to save CPU time. Indeed, we can
compute
less often the optical coefficients since they vary slowly in time.
The Planck function, which varies more quickly, can easily be computed more
often, because it is just a polynomial approximation.
This new method is less costly and more accurate.
For details, report to ESA Contract 11369/95/NL/JG (Work Package VI:
Technical Report - New model radiation schemes, parameterization and
validation).
in the original code developed by Fouquart and Bonel (1980), the upward and downward fluxes are obtained from the reflectance and transmittance of the layers. The interaction between gaseous absorption and scattering (by dust, molecules or clouds) is introduced using the photon path distribution method. At this stage, only absorption and scattering by dust (already present in the version of the code used at the ECMWF) is included in the Martian version although absorption by the near infra-red bands of carbon dioxide may become non negligible for very non-dusty conditions. The transmittance and reflectance of the layers are computed using the Delta-approximation to account for the strong asymmetry of the aerosol phase function.